ISSN: 2278-5183

www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

Comparative Analysis of Multipath Routing Algorithms for Mobile Ad-hoc Networks

Divya Sharma1 & Ishani Mishra2
1&2
Asst. Professors, New Horizon College of Engineering, Bangalore, India
ABSTRACTA MANET is an interconnection of mobile devices by wireless links, which forms a dynamic
topology. Routing protocols play a vital role in transmission of data across the network. Multipath routing
allows the establishment of multiple paths between a pair of source and destination node in mobile ad hoc
network. It is typically proposed in order to increase the reliability of data transmission or to provide load
balancing and has received more and more attentions. In this paper we have compared & evaluated the
performance of three multipath algorithms namely AOMDV, MSR and MDART on different parameters such
as throughput and delay.
KEYWORDS : MANET, Multipath Routing, AOMDV, MSR, MDART and CBR.
1. INTRODUCTION
Mobile ad hoc network (MANET) consists of several wireless mobile nodes which dynamically exchange
data among themselves without the reliance on a fixed base station or a wired backbone network. Due to the
limited transmission power, multiple hops are usually needed for a node to exchange information with any
other node in the network. So routing discovery and maintenance is crucial issues in MANET.
Routing in MANETs must take into consideration their important characteristics such as node mobility. Work
on single path (or unipath) routing in MANETs has been proposed in [1] [2]. In this paper, we specifically
compare few characteristics of some multipath routing protocols in MANETs. Multipath routing allows the
establishment of multiple paths between a single source and single destination node. Multipath routing is
typically proposed in order to increase the reliability of data transmission (i.e., fault tolerance) or to provide
load balancing. Load balancing is of special importance in MANETs because of the limited bandwidth
between the nodes.
The most popular on-demand routing protocol, Adhoc On-demand Multipath Distance Vector (AOMDV)
routing protocol [3] is an improvement of Ad-hoc On-demand Routing Protocol (AODV).
AOMDV discovers multiple paths between a source and destination to provide efficient fault

tolerance by providing quicker and more efficient recovery from route failures in a dynamic network. As
AOMDV discovering multiple paths in a single route discovery attempt, new route needs to be discovered
only when all paths fail. This reduces not merely the route discovery latency but the routing overheads also.
M-DART is an enhancement of shortest path routing protocol known as Dynamic Address Routing (DART)
[4]. M-DART discovers and stores multiple paths to the destination in the routing table. M-DART extends the
DART protocol to discover multiple routes between the source and the destination. In such a way, M-DART
is able to improve the tolerance of a tree-based address space against mobility as well as channel
impairments.
MSR proposed in [5] is a dynamic multipath source routing (MSR) protocol with QoS support which shows
higher performance compared to the traditional unipath DSR protocol.
2. AN OVERVIEW OF MULTIPATH
PROTOCOLS
2.1 MULTIPATH ROUTING
Multipath routing aims to find multiple routes between source and destination node. These multiple paths
between source and destination node pairs can be used to compensate for the dynamic and unpredictable
nature of MANET, and support QoS. Multipath based routing protocols can discover node disjoint, link
disjoint, or non-disjoint routes. Node disjoint routes, also known as totally disjoint routes, have no nodes or
links in common. Link disjoint routes have no links in common, but may have nodes in common. Non-disjoint
routes can have nodes and links in common [5]. Non-disjoint routes may have lower aggregate resources than
disjoint routes, because non-disjoint routes share links or nodes. While the advantage of non-disjoint routes is
that they can be more easily discovered, since there are no restrictions that require the routes to be node or link
disjoint. In QoS routing, only a subset of paths that satisfies the QoS requirement will be selected
8|P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

2.1 AOMDV :
AOMDV [6], [4] is a multi-path routing protocol. It is an extension to AODV and also provides two main
services i.e. route discovery and maintenance. Unlike AODV, every RREP is being considered by the source
node and thus multiple paths can be discovered in one route discovery. Being the hop-by hop routing
protocol, the intermediate node can maintain multiple path entries in their respective routing table. To
discover distinct paths, AOMDV suppresses duplicate route requests (RREQs) at intermediate nodes. Such
suppression comes in two different variations, resulting in either node (illustrated in Fig. 1 (a)) or link
(illustrated in Fig. 1(b)) disjoint.
AOMDV can be configured to either discover the link (no common link between any given pair of nodes) or
node (in addition to link disjoint, common intermediate nodes are also excluded between any given pair of
nodes) disjoints paths. Disjoint alternate paths are a good choice than overlapping alternate paths, as the
probability of their interrelated and concurrent failure is smaller. This property can be helpful in an adversarial
environment where malicious activity can also cause additional link failure. Finding a disjoint path is quite
straightforward in source routing (as every node maintain complete path information for every path), but hopby-hop routing i.e. AOMDV is considered more efficient in terms of creating less overhead Number of paths
in any given source and destination is directly proportional to the number of nodes in entire network.
AOMDV works more efficiently in dense and heavy networks.

Figure1. AOMDV Multipath protocol

2.3 MULTIPATH SOURCE ROUTING:

MSR discussed in [5] consists of three phases: routing discovery, routing maintenance and traffic allocation.
MSR will send RREQ to discover routing which is similar to DSR. The MSR RREQ mainly includes the
following fields: SourceID and RREQID, which are used to uniquely identify a QoS routing request.
Routing_list, which is used to keep track of the RREQ traveling along nodes and so that the destination node
can select node disjoint multipath routing.
Hmax, which is used to avoid overhead brought by routing with too many hops.
Bmin and Rmin, which denote the minimum bandwidth and reliability requirements.
Routing discovery can be triggered upon failure of the route and initiating new QoS request. The source node
firstly checks whether it has the routing information to the destination node. If not, it begins to broadcast
RREQs to its neighborhoods. Once intermediate nodes receive this RREQ, Unlike DSR, they do not keep a
route cache and therefore, do not reply to RREQs. This helps to allow the destination node to receive all the
routes so that it can select the maximally disjoint paths. Duplicate RREQs are not necessarily discarded.
Instead, intermediate nodes forward RREQs that are received through a different incoming link, and whose
hop count is not larger than the previously received RREQs. The process on each intermediate node can be
described as follows:
Step 1 If current node itself is within the Routing_list recorded by RREQ, it will discard the RREQ because of
the routing loop. Otherwise, goto step 2;
Step 2 If the tuple (SourceID, RREQID) of RREQ is not included in the routing table, which means the
current node is the first time to receive this RREQ, it calculates the corresponding value of the bandwidth
according to Eq.(1) and (2). If the value is not less than Bmin, the RREQ will be discarded. Otherwise, goto
step 3;
Step 3 Append the value of bandwidth to the
9|P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

corresponding fields of the RREQ. Then the

RREQ will be continually forwarded. Goto step 1.
The destination sends an RREP for the first RREQ it receives, which represents the shortest delay path. The
destination then waits to receive more RREQs. From the received RREQs, the destination node can construct
a certain topology for network and the path that is maximally disjoint from the shortest delay path is selected.
If more than one maximally disjoint path exists, the shortest hop path is selected. If more than one shortest
hop path exists, the path whose RREQ was received first is selected. The destination then sends an RREP for
the selected RREQ. The process can be described as follows:
Step 1 Initialize the maximal number of routes N and Hmax;
Step 2 Calculate end-to-end reliability according to Eq.(3), and sort the sequence(Routing_list1, Routing_ list2
Routing_list n) in descending order based on the reliability values. Routing_listi denotes the ith Routing_list
in the corresponding field in RREQ. RREQ recording with Routing_list1 will be added into the responding
buffer;
Step 3 From Routing_list2 to Routing_listn, find all Routing_listi (i=2, 3, , n) which disjoint with
Routing_list1 [7]. If the number of the multipath routes founded is no more than N and the hops of each route
is less than Hmax, the corresponding RREQ will be added into the responding buffer and the count of routes
will be increased by 1;
Step 4 If the iteration finished or the routing number exceeds N, terminate the process. Otherwise, goto step3.
2.3 MDART:
The protocol, namely the multi-path dynamic address routing (M-DART) [9], is based on a prominent DHT
based shortest-path routing protocol known as DART [4,8 ].
A. Address space: The network addresses are
strings of l bits, thus the address space structure can be represented as a complete binary tree of l + 1 levels,
that is a binary tree in which every vertex has zero or two children and all leaves are at the same level (Figure
2a). In the tree structure, each leaf is associated with a network address, and an inner vertex of level k, namely
a level-k subtree, represents a set of leaves (that is a set of network addresses) sharing an address prefix of l
k bits. For example, with reference to Figure 2a, the vertex with the label 01X is a level-1 subtree and
represents the leaves 010 and 011. Let us define level-k sibling of a leaf as the level-k subtree which shares
the same parent with the level-k subtree the leaf belongs to. Therefore, each address has l siblings at all and
each other address belongs to one and only one of these siblings. Referring to the previous example, the vertex
with the label1XXis the level-2 sibling of the address 000, and the address 100 belongs only to this sibling. In
Figure 2b, the address space is alternatively represented as an overlay network built upon the underlying
physical topology. Its tree-based structure offers simple and manageable procedures for address allocation,
avoiding to rely on inefficient mechanisms like flooding.
B. Route discovery and packet forwarding:
Each node maintains a routing table composed by l sections, one for each sibling, and the kth section stores
the path toward a node belonging to the level-k sibling. Each section stores five fields: the sibling to which the
entry refers to, the next hop, the cost needed to reach a node belonging to that sibling using the next hop as
forwarder, the network id used for address validation, and the route log used by the loop avoidance
mechanism.. The table has three sections: the first stores the best route, according to a certain metric, toward
the node 001, the second toward a node belonging to the sibling 01X, and the last toward nodes belonging to
the sibling 1XX.The routing state information maintained by each node is kept consistent through the network
by means of periodic routing updates exchanged by neighbour nodes. Each routing update stores l entries, and
each entry is composed by four fields: the sibling id, the cost, the network id, and the route log.
The packet forwarding process exploits a hop-by-hop routing based on the network addresses and it is
summarized
by Algorithm 1. To route a packet, a node compares its network address with the destination one, one bit at a
time starting with the most significant (left-side) bit, say the l th. If the I th bit is different, the node forwards
the packet towards one the route stored in the I th section. With reference to the previous example, if the node
000 has to send a packet to the node with the address 101, then it will forward the packet to the next hop
stored in the third section (i.e., the node 010).

10 | P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

Figure 2. Relationship between the Address Space & Physical Topology

3. PERFORMANCE EVALUATION
3.1 SIMULATION PARAMETERS
For simulation study, we develop extensive routing program on MATLAB7.12.0. The networks with no. Of
nodes varying from 10 to 60 are randomly generated within a 1000 X 1000 square region. The IEEE 802.11
MAC protocol is used in the network. A random way-point is selected as movement model and Constant Bit
Rate (CBR) is used to send data. The qualities of Constant Bit Rate (CBR) traffic pattern [6] are i)
unreliable: since it has no connection establishment phase, there is no guarantee that the data is transmitted to
the destination, ii) unidirectional: there will be no acknowledgment from destination for confirming the data
transmission and iii) predictable: fixed packet size, fixed interval between packets, and fixed stream duration.
In all cases, our results are based on the performance of 15 randomly generated networks.
TABLE 1. SIMULATION PARAMETERS

Parameter
Area
Number of Nodes
Traffic Type
Simulation Time
Routing Protocols

Value
1000m X 1000m
10,20,30,40,50,60
CBR
100s
AOMDV,MSR,MDART

3.2 SIMULATION RESULTS

A. PACKET DELIVERY RATIO:
As shown in figure 3(see figure 3 a,b,c), for small no of nodes, the packet delivery ratio for MSR is slightly
greater than AOMDV and MDART and However, as the no. of nodes increases, the burden of intermediate
nodes will be aggravated and will result in failure of resourse reserving. The performance of all three
degrades, however MSR still prove to have better throughput. AOMDV and MDART show almost the same
packet delivery ratio.
B.Average End to End Delay:
As shown in figure 4(see figure 4 a,b,c), For small no. of nodes, AOMDV and MDART shows approximately
same end to end delay. As the no. of nodes increases, end to end delay of MSR grows linearly and is higher
than AOMDV and MDART.

11 | P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

Fig 3a) Results for AOMDV

Fig 3b) Results for MSR

Fig 3c) Results for MDART

Figure 3: Packet Delivery Ratio vs Traffic Load

Fig 4a) Results for AOMDV

12 | P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

Fig 4b) Results for MSR

Fig 4c) Results for MDART

Figure 3: End To End Delay vs Traffic Load

4. CONCLUSION
Multipath routing supports scalability in various wireless networks. As per our simulation results, the packet
delivery ratio for MSR is slightly greater than AOMDV and MDART. With the increase in traffic load, the
average end to end delay for MSR gets increased linearly and is higher than AOMDV
and MDART. In future work, AOMDV & MDART can also be modified with QOS parameters and further be
compared with QOS MSR.
REFERENCES

[1]. Perkins, C.E., Royer, E.M.: Ad-hoc On-Demand Distance Vector Routing. Proceedings of the 2nd IEEE
Workshop on Mobile Computing Systems and Applications
(1999)
[2]. Johnson, D.B., Maltz, D.A.: Dynamic Source Routing in Ad Hoc Wireless Networks.
Mobile Computing. (1996) 153-181
[3]M.Z.Oo,M.Othman,Performance Comparisons of AOMDV and OLSR Routing Protocols for Mobile
AdHocNetwork, Second International Conference on Computer Engineering and Applications, 2010.
[4] J. Eriksson, M. Faloutsos and S. Krishnamurthy.DART: Dynamic Address Routing for Scalable Ad Hoc
and Mesh Networks. In IEEE- ACM Transactions on
Networking, vol. 15, no. 1, April 2007, pp. 119-132.
[5] Multipath Routing for Mobile Ad Hoc Network ISBN 978-952-5726-02-2 (Print), 978-952-5726-03-9
(CD-ROM) Proceedings of the 2009 International Symposium on Information Processing (ISIP09)
Huangshan, P. R. China, August 21-23, 2009, pp. 237-240

13 | P a g e

www.cirworld.com

ISSN: 2278-5183
www.ijcdsonline.com

International Journal of Computers and Distributed Systems

Vol. No.3, Issue III, Aug-Sep 2013

[6] Elizabeth M. Royer, Charles E. Perkins An Implementation study of AODV routing protocol IEEE, 07803-6596- 8/00,2000.
[7] D.Ganesan and R. Govindan, Highly-Resilient,Energy- Effient Multipath Routing in Wireless Sensor
Networks, Mobile Computing and Communications Review, 2001, pp.11-25.
[8] Caleffi, M., Ferraiuolo, G., and Paura, L., Augmented Tree based Routing Protocol for Scalable Ad Hoc
Networks, in MHWMN '07: Proceedings of the Third IEEE International
Workshop on Heterogeneous Multi-Hop Wireless and Mobile Networks, 2007..
[9] Avinash giri, Jitendra Prithviraj & Ashok Verma Analysis of unipath and multipath routing protocols in
mobile adhoc networks , International Journal of Smart Sensors and Ad Hoc Networks (IJSSAN), ISSN No.
2248-9738 (Print), Vol-2, Iss-1,2, 2012

14 | P a g e

www.cirworld.com